ChatGPT drops Advanced and Standard voice on December 11 for GPT-Live
ChatGPT will stop offering Advanced voice and Standard voice on December 11, 2026. OpenAI announced the change on November 10 in a post on its official OpenAI Newsroom account, saying it is retiring both options from ChatGPT Voice and that every voice conversation will run on GPT-Live, the full-duplex voice model it introduced in ChatGPT in July. Users who still have one of the two older modes selected will be moved to Live voice automatically, and both entries will disappear from the voice choices under Settings.
The company gave two reasons in its statement. Voice usage in ChatGPT has nearly doubled since GPT-Live arrived, according to OpenAI, and virtually all people who use voice in ChatGPT already talk to the new model. OpenAI also said the change will allow it to simplify the voice experience and focus on improving GPT-Live, and that feedback from people who preferred the older modes will be used to work on response depth and detail, continuity across conversations, and personalization.
The retirement comes with two exceptions that keep Advanced voice alive for a while longer. Live video and screen sharing will continue to run on Advanced voice until GPT-Live supports them, and OpenAI has not given a date for that. Voice conversations with custom GPTs will also keep using Advanced voice until December 11, the same day custom GPTs reach their own scheduled retirement for most users. Users therefore have 31 days of notice, counted from the November 10 announcement.
··········
WHAT OPENAI ANNOUNCED ABOUT CHATGPT VOICE ON NOVEMBER 10.
The announcement was short and was made on the company's official account rather than in a long blog post. OpenAI said it is retiring Advanced and Standard voice from ChatGPT Voice on December 11, 2026, recalled that it introduced GPT-Live in July as its latest voice model combining natural conversation with frontier intelligence, and attributed the decision to the shift in usage that followed. The company said conversations started in Advanced or Standard voice will remain in ChatGPT history and can be continued afterwards, either by typing or by speaking, with spoken replies handled by GPT-Live.
OpenAI's statement framed the move as a simplification. Until now ChatGPT has kept three voice paths side by side: GPT-Live as the default for most users, Advanced voice as the earlier speech-to-speech mode, and Standard voice as the oldest option, which many long-time users kept because it read out the answers of the text model selected in the conversation. Running three systems in parallel meant three sets of behaviour, voices and limitations to document and maintain. After December 11 only one of them will be available for ordinary voice chats.
The company has not published a dedicated blog post or a full feature matrix for the change at the time of writing. The details below come from OpenAI's official post and from the statements the company gave alongside it, and they are attributed accordingly.
··········
CHATGPT VOICE BEFORE AND AFTER DECEMBER 11.
The most useful way to read the change is mode by mode. The table summarises what each part of ChatGPT Voice does today and what OpenAI says will happen after the retirement date.
........
ChatGPT voice feature | Until December 10, 2026 | From December 11, 2026 |
GPT-Live (Live voice) | Available, default for most users | Only voice option for chats |
Advanced voice | Selectable in Settings > Voice | Removed from Settings |
Standard voice | Selectable in Settings > Voice | Removed from Settings |
Users on an older mode | Keep their selection | Moved to Live voice automatically |
Old voice conversations | In chat history | Stay in history, continue by text or GPT-Live |
Custom GPT voice chats | Run on Advanced voice | Custom GPTs reach scheduled retirement |
Live video and screen sharing | Run on Advanced voice | Stay on Advanced voice until GPT-Live supports them |
........
Two rows deserve attention. Video and screen sharing are the reason Advanced voice is not being switched off completely: the older model keeps powering those features for eligible users, and OpenAI has tied their move to the moment GPT-Live gains equivalent capabilities. The custom GPT row is connected to a separate program in which OpenAI is helping creators move custom GPTs to plugins; FedRAMP and some Enterprise workspaces have a later retirement date for custom GPTs, so voice behaviour inside those organisations may differ from the consumer plans until their own deadline.
··········
HOW GPT-LIVE DIFFERS FROM ADVANCED AND STANDARD VOICE.
GPT-Live is a full-duplex model, which means it can listen and speak at the same time instead of waiting for the user to finish a turn. OpenAI describes it as its model for natural, expressive voice conversations with smooth interruption handling. The model can give short acknowledgments while the user is still talking and can be interrupted mid-sentence without losing the thread, behaviour that turn-based voice systems handle poorly.
The second design choice is delegation. GPT-Live handles the conversation itself but can hand deeper reasoning, tool use and longer tasks to other OpenAI models running in the background. In ChatGPT this lets a voice chat trigger work that needs a stronger text model, and OpenAI cites Codex and ChatGPT Work as examples of where that delegation already happens. Standard voice approached the problem from the opposite side: it converted speech to text, sent it to the text model chosen in the chat, and read the written answer aloud. That made it slower and less conversational, but it gave users the full depth and formatting logic of the text model, which explains why a vocal group of Plus and Pro subscribers asked OpenAI to keep it.
OpenAI has updated the model once since launch. GPT-Live-1.1 began rolling out to all ChatGPT users on October 15, 2026, with 13 new voices, better accents and stronger performance in French, German, Hindi, Mexican Spanish, Korean, Japanese and other languages, together with more natural delegation and improved steerability, according to the company's announcement. The three areas OpenAI now promises to improve using feedback from users of the older modes, namely response detail, continuity across conversations and personalization, match the complaints users have raised most often about Live voice compared with Standard voice.
··········
WHAT CHANGES FOR CHATGPT USERS, CUSTOM GPTS AND VIDEO CALLS.
For most people nothing needs to be done. Anyone already using Live voice will see no difference on December 11, and anyone still on Advanced or Standard voice will be switched by ChatGPT without a manual migration. The practical steps concern the users who chose an older mode deliberately.
People who relied on Standard voice to hear the full answer of a specific text model can still get that result in two ways after the retirement. They can ask GPT-Live for detailed answers and let it delegate the work to a background model, or they can type the request and use ChatGPT's read-aloud option on the written reply, which keeps the full text answer in the conversation. Users who want to compare behaviour before the deadline have until December 10 to test both paths with their own prompts. Our guide to how ChatGPT real-time voice works and how to fix common issues explains the settings involved, and our earlier report on GPT-Live in ChatGPT Voice and its plan limits covers the model's July launch.
Users of live video and screen sharing keep those features on Advanced voice for now, so a camera or screen session started after December 11 may still sound different from an ordinary voice chat. Creators and teams that built voice workflows on custom GPTs should treat December 11 as a hard date, because OpenAI ties the voice change to the custom GPT retirement and recommends moving those workflows to plugins. Enterprise and FedRAMP administrators should check the retirement date that applies to their own workspace, since OpenAI has set later custom GPT deadlines for some of them.
Existing voice conversations are preserved in chat history. A long-running chat that started in Standard voice can be reopened after the retirement and continued by text, and any new spoken reply in that chat will come from GPT-Live.
··········
GPT-LIVE IN THE OPENAI API: WHAT DEVELOPERS ALREADY PAY.
The retirement concerns the ChatGPT app, and OpenAI has not announced any change to its developer voice models in the same statement. GPT-Live has been available to developers as gpt-live-1 since September 10, 2026, and its official model page lists the terms below.
........
GPT-Live-1 in the OpenAI API | Official value |
Voice session price | $0.05 per minute, billed per second |
Backend model and tools | Billed separately at normal rates |
Input modalities | Audio, text |
Output modalities | Audio, text |
Image and video input | Not supported |
Function calling | Supported |
Concurrent sessions | Build 50, Launch 300, Grow 500, Free tier not supported |
Knowledge cutoff | July 31, 2025 |
........
The pricing structure follows the same split as the consumer product. The per-minute fee pays for the conversational front end, while reasoning and tool calls are passed to a backend text model, such as GPT-6 Astra or a third-party model, and billed at that model's rates. Our report on GPT-Live-1 in the API and its telephony pricing covers the developer launch in detail. The lack of image and video input in the API model page is consistent with the ChatGPT exception for live video and screen sharing, which still depends on Advanced voice.
··········
CHATGPT VOICE RETIREMENT: CONFIRMED FACTS AND OPEN QUESTIONS.
OpenAI has confirmed the date, the scope and the migration path. Advanced and Standard voice leave ChatGPT on December 11, 2026; users on either mode move to Live voice automatically; earlier voice conversations stay in history; custom GPT voice chats run on Advanced voice until the same date; and live video and screen sharing remain on Advanced voice until GPT-Live supports them. The company's stated rationale rests on its own usage figures, which say voice use has nearly doubled since July and that virtually all voice users already rely on GPT-Live.
Several points remain undocumented. OpenAI has not said when GPT-Live will handle video and screen sharing, which means Advanced voice has no announced end date for those features. The company has not published plan-by-plan voice limits alongside the change or said whether any plan, including Pro, Business or Enterprise, can keep an older mode as a legacy option. It has also not quantified the improvements it plans for response depth, continuity and personalization. Until a help-centre article with those details appears, the reliable facts are the December 11 date, the automatic switch to Live voice and the two temporary exceptions for video, screen sharing and custom GPTs.
··········
FOLLOW US FOR MORE.
·····
DATA STUDIOS
·····
datastudios.org
